A Probabilistic Reasoning-based Multi-agent Formalism for Smart Environment

Abstract

Recent years have witnessed rapid advances in ubiquitous computing paradigms that significantly influence human life through enhanced comfort, security, and automation. Although smart devices in context-aware environments demonstrate adaptive behavior, their interaction with users often produces uncertainty and inconsistency, particularly in safety-critical domains. To address these challenges, this paper proposes an ontology-driven, probabilistic reasoning-based multi-agent formalism that unifies semantic expressiveness with probabilistic inference to support reliable decision-making under uncertainty. The formalism was instantiated in a smart fridge case study, where OWL~2~RL ontologies and SWRL rules modeled contextual knowledge, autonomous agents dynamically updated beliefs, and probabilistic reasoning guided replenishment and safety decisions. Formal verification using the UPPAAL model checker demonstrated that the system satisfies the essential correctness properties, including safety, liveness, reachability, and freedom from deadlock. The experimental results confirmed that the architecture prevents unsafe actions, guarantees process progression, and is effective in adapting to dynamic contextual changes. Beyond the smart fridge scenario, the proposed framework is broadly applicable to healthcare, transportation, and energy management systems. Future work will extend the approach toward predictive probabilistic reasoning and large-scale real-world deployment.